基于ICT切片图像的轮廓特征识别和曲线拟合  

Contour feature recognition and curve fitting based on ICT slicing images

作　　者：方黎勇[1] 戴明宏[2] 雷华堂[1] 李柏林[1] 

机构地区：[1]西南交通大学机械工程学院,成都610031 [2]郑州铁路职业技术学院,郑州450052

出　　处：《机械设计与制造》2010年第1期211-213,共3页Machinery Design & Manufacture

基　　金：国家863计划项目(2003AA4Z3200)

摘　　要：在基于ICT切片图像的逆向工程中,切片图像的轮廓特征处理技术具有重要的意义,而轮廓整体和分段特征的识别、曲线拟合是其首要的环节。提出一种新的轮廓处理方法:首先采用轮廓矩不变量的原理识别出所有闭合轮廓中的圆和正多边形整体特征;再根据曲率对剩下的轮廓进行特征点提取,确定出每段轮廓对应的曲线类型;最后采用最小二乘法对识别出的整体和分段特征进行曲线拟合。试验表明:算法自动化程度较高,而且处理结果准确可靠,精度较高。The processing technology of contour feature is an important part of the reverse moaetmg based on ICT （Industrial Computerized Tomography） slicing images. Especially,the whole and segmenta- tion feature recognition and curve fitting are primary part in the process of reverse modeling based on feature. Therefore ,a new algorithm for the whole and segmentation features processing is proposed. In this algorithm,the theory of the contour-based moment invariants is used to recognition the closed circle and regular polygon feature firstly. Then,the feature points recognition and curve type determination of segmented data points are decided according to the discrete curvature aiming to the residual closed contours. Finally,the least square method is used to fit the curve of the whole and segmentation features. The experimental results showed that the proposed algorithm is effective and accurate ,and have a high automation.

关 键 词：ICT切片图像 轮廓矩不变量 特征识别 曲线拟合 最小二乘法 

分 类 号：TH16[机械工程—机械制造及自动化] TP391.4[自动化与计算机技术—计算机应用技术]